Fixing the NBA Draft

The NBA recently voted to change the way it decides the order teams draft college players. If you aren’t familiar with basketball, the current system determines the order by assigning different probabilities of drafting in each position to each of the 30 teams based on the standings. For example, under the old system, the team that finished last had a 25% chance of receiving the top overall pick, while the 14th worst had only a 0.5% chance to land in the top spot (playoff teams are not part of the lottery – they just pick in reverse order of standings).

This system is distinct from other sports like the NFL, which deterministically sets the draft order as the reverse of the standings. The worst team always picks first in football. The reason the NBA does not follow the NFL is to discourage “tanking,” which is when one team attempts to lose on purpose to get the number one pick and improve their team for the future. A new proposal that was just approved by a majority of owners punishes tankers even more, reducing the odds of the worst team getting the first pick to 14% and giving the exact same 14% chance to the second and third worst teams.

At first, it might seem like the lottery system accomplishes its goal of reducing the incentives to tank. The benefit of coming in last place is obviously higher if you have a 100% chance of getting the first pick rather than a 25% chance or a 14% chance.

But there’s something wrong with this logic. The decision to tank or not does not depend on the overall benefit of coming in last, but rather its relative benefit compared to any other strategy. In other words, the only reason a team wouldn’t tank is if its benefit of playing hard every game outweighs the benefit of losing and moving down the standings. It seems clear to me that under any system that gives any draft advantage to the worst teams will always encourage tanking.

There is simply no benefit at all to being the 24th or 23rd best team in the NBA. If your team cannot realistically compete for a title, you are always better off being dead last than somewhere in the middle. With the new system, that calculation changes so that you become indifferent between any of the bottom 3 spots, but does anybody really care much if a team only has to tank to 28th instead of 30th? They are still better off losing as many games as possible.

This flaw in any lottery system has led to even more radical proposals that decouple draft order and standings completely. The most famous of these is “the wheel,” which would replace the lottery entirely with a draft order set years in advance. Each team would pick in each of the 30 draft slots exactly once every 30 years. And they would know exactly when. No randomness. No relation to the standings at all.

The virtue of this system is that it removes all incentive to tank. If your draft order is unconnected to your record, you might as well do your best to win. This feature has given it a large following of NBA fans hoping for more competition in the league and it has been seriously considered as an alternative to the lottery.

I think it’s a terrible idea. By removing the link between record and draft order, the wheel solves the tanking problem. But it deepens another major issue with the NBA: how do bad teams get better? And what happens when a team like the Warriors ends up with the first pick in the draft? Imagine the current Warriors roster plus Markelle Fultz and you can immediately see why the wheel can end up producing some incredibly undesirable results.

In the article I linked above, Zach Lowe acknowledges both of these issues, but writes them off by arguing that they would be a part of any draft system that offers a chance for good teams to get good picks. And he’s absolutely right. Any measure that discourages tanking will necessarily make it harder for bad teams to get better. And any attempt to give some advantage to bad teams will always encourage tanking. There is no perfect system.

My proposal is a bit different. Rather than try to fix the draft, fix the system that makes tanking one of the few ways for a team in a bad situation to improve. Tanking is not the the root of the problem. The issue is that teams like the Warriors can have 4 superstar players on the roster, making it nearly impossible for teams with less talent to compete. What’s the point of trying to win when you know you won’t? Even a tiny advantage in the draft is enough for any team to see some value in tanking when their probability of beating the teams at the top falls close to zero.

We can solve that problem in a much easier way. Remove the cap on player salaries. Let LeBron and Durant make $50 million a year. The market would make it impossible for the Warriors to have multiple top 5 players. Somebody would offer enough that one of them would want to leave. And let bad teams tank. Get rid of the lottery. The worst team gets the first pick. Would tanking increase? Possibly, but so would parity in the league. Nobody watches bad teams anyway. At least this system would give them a path to being good again.

Keynesian Economics and Monetary Disequilibrium

What is Keynesian Economics about? Even if you’ve never taken an economics class you might still have some idea. The 2008 stimulus package was frequently referred to as a Keynesian policy. Government intervention and Keynesian economics often go hand and hand. If you have taken a macroeconomics class you might have an even deeper knowledge of Keynesian economics. Maybe you know about the multiplier, the Keynesian cross, the IS-LM model. And all of those are certainly related to Keynesian economics, but none really capture the heart of Keynes’s contribution.

Part of the trouble with people’s understanding of Keynes’s work is that secondary sources frequently distort what he actually said. For example, if you read Mankiw’s intermediate macroeconomics textbook you will come away with the impression that Keynesian economics is about sticky prices. We get unemployment because wages don’t adjust downward quickly enough. So called “New Keynesian” models, the modern analogue to IS-LM are also predicated on slower than optimal price adjustments to shocks. It is true that Keynes assumed sticky prices for part of his analysis, but he was careful to emphasize that “The essential character of the argument is precisely the same whether or not money-wages, etc., are liable to change” (General Theory Ch. 3).

So if not sticky prices or wages, what is it that causes unemployment in Keynes’s world? In my reading, Keynes’s story is all about monetary disequilibrium and a failure to coordinate savings and investment in a monetary economy.

Keynes famously stated that his theory refuted “Say’s Law,” which can be simply stated as “supply creates its own demand.” Now, whether or not that’s what Say actually said remains a point of contention even today, but  it’s that formulation that Keynes attempts to refute, which I think is worth exploring on it’s own. What Keynes really wants to do is draw a distinction between a barter economy and a monetary economy. I recently taught an intermediate economics class and I developed a simple example that I think helps illustrate the main points.

Let’s start with an economy with no money and only two goods, apples and bananas. There are two people in the economy. Person A can only produce apples and person B can only produce bananas. For simplicity, I will assume a fixed price ratio of 2:1. One banana is worth as much as two apples. The story gets more complicated if we allow this price to change, but I don’t think the main implications would be any different. In this economy with no money, the only trade that can occur is bananas for apples. If A wants to demand 10 bananas from B, he needs to produce 20 apples. It is in this sense that supply creates its own demand. Without money, demand for one good is precisely supply of another. The diagram below illustrates what is happening (I arbitrarily fixed prices in dollar terms, but remember there is still no money. Dollars operate only as a unit of account)

In this kind of barter setup, Say’s law is trivially true. Any demand for bananas is supply of apples so supply creates its own demand. In this case, demand for bananas is $10(20) = 200 and supply is $20(10), which are obviously equal. However, when we start to add in money, the relationship between supply and demand is not as clear.

For a second example we will assume now that bananas and apples cannot be directly traded. Instead, each good will have to be sold for dollars, which can then be used to purchase the other good. Of course, we could still have the exact same situation as above. If all transactions happen instantaneously and every time an apple is produced it is immediately sold and the proceeds immediately used to purchase an apple. However, now there is another possibility.

Imagine that for some reason the apple producer wants to consume more bananas tomorrow than today and decides to save in dollars (I assume apples cannot be saved directly – this might be a strong assumption but it will make sense later). He still wants to consume 10 bananas (so he needs to produce 20 apples), but he also wants to save $100, so he produces an additional 10 apples (30 apples total). But what if the banana producer still only wants to buy 20 apples. He doesn’t want to pay an extra $100 for the additional 10 apples that are being produced.

If you’re an economist the first question that comes to your mind should be why the price doesn’t just adjust. If demand for apples is less than supply for apples, the price of apples should fall until the market is equilibrated. So in this case, apples could fall in price to $6.67 so that 10 bananas buy 30 apples and supply equals demand again. Except then we run into a problem. After the price adjustment producer A still isn’t happy. He didn’t get to save his $100. So really this can’t be an equilibrium at all (it’s possible that the change in prices would also affect his desired saving, but as long as it’s still positive it’s still not an equilibrium).

The problem with the logic above is that we are still trying to think of the economy as a barter economy with one market (trade between apples and bananas), when it is actually a monetary economy with 2 markets (money for apples and money for bananas). Finding a single equilibrium price for two markets is not enough. We need equilibrium in both markets. What we have in the example above is an excess supply of apples and an excess demand for money. The banana producer can’t supply additional money, so he is unable to help return to equilibrium. So who can help? Who supplies money? The Fed! If the Fed simply prints money to buy the excess apples we are back to equilibrium at the same prices as before.

So what does any of this have to do with unemployment? Let’s change the example slightly. Instead of apples, assume now that A provides labor. They work for B to produce bananas. If this is a barter economy then their wage is paid in bananas (I still fix arbitrary dollar values) and demand and supply are always equal as above:

But what if the worker wants to save? He obviously can’t save labor directly (which is why I assumed no saving of apples above) and I will assume he doesn’t have the storage to save bananas either. Instead, he will try to save in dollars by buying fewer bananas (but working the same amount). If he wants to save $100 the picture becomes:

Since the worker still worked 20 hours, production of bananas didn’t fall, but demand for bananas did. The firm is forced to put the extra bananas into its inventories. Is this a problem? Maybe not. If the firm realizes that the worker will use his savings to purchase more bananas in the future, they are happy to increase their investment now in order to produce more bananas next period (an equivalent story could be told where they are building machines to increase production, but the inventory version is the simplest I think). We could then imagine a second period of this economy where the worker uses his savings from the first period to buy the bananas from the inventory in the second.

This story gives us a nice equilibrium outcome with no unemployment. The worker works his desired amount in each period, saving $100 in the first to buy five additional bananas in the second. The banana producer invests in 5 extra bananas to prepare for the increase in demand in the second period. Saving and investment are perfectly coordinated. Say’s Law holds.

But it is easy to imagine another equilibrium. What if the firm does not realize demand for its product will increase next period? All it knows is that consumers want to buy 5 fewer bananas right now. In this simple example where there is only one option for the worker to spend his money that’s hard to believe, but in the real economy with thousands of firms and millions of products, an individual producer has no idea whether increased savings will translate to future demand for its own product. If demand falls today, they might predict lower demand tomorrow as well. In this case they will not want to increase investment today. Instead they simply cut production. So we have a worse equilibrium that looks like

Here the firm doesn’t expect the decline in demand today to translate into an increase in demand tomorrow. They see demand for 5 bananas today so they only hire the worker for long enough to produce those 5. The worker still wants to work 20 hours to save an additional $100, but nobody will hire him so he only works 10 and is therefore underemployed (if we think about this as representing many people we would have some employed and some unemployed). Keynes referred to this situation as involuntary unemployment. People want to work at the prevailing wage, but since firms don’t expect their consumption of their product to compensate the cost of their wages, they don’t want to hire.

There are two interesting points here. First, note that wage cuts will not help. The worker wants to save regardless of his wage. Cutting it will only make him want to work more, which actually makes the problem even worse. Second, the firm’s prediction actually comes true. Since the worker was actually unable to save anything in the first period, his demand for bananas actually won’t increase in the future either. By expecting lower demand tomorrow, the firm actually caused that future to be realized. In this way, expectations are self-fulfilling and we get multiple equilibria.

This example is highly stylized, but I think it demonstrates Keynes’s main point. When somebody wants to save, there is no magical process that instantly transforms their saving into investment. Investment decisions are driven primarily by firms’ expectations about demand for their own products. Without perfect foresight, they must rely on cruder measures of prediction (like animal spirits). If they don’t expect increased saving to translate into future demand, we get unemployment.

I will have more to say on Keynesian economics in at least one future post (including some criticism of Keynes), but this is already getting long so I will stop here for now.


Seeing Like a State

I recently finished reading Seeing Like a State, an interesting book by James Scott, a political scientist at Yale. Scott argues that many attempts to coordinate and control from the top down necessarily leave out many important details that are essential to the workings of an organically developed process. In his words “Designed or planned social order is necessarily schematic; it always ignores essential features of any real, functioning social order.” His thesis is essentially a Hayekian one. Because they can never fully collect the local knowledge of individuals, state programs often forget important features of society, in some cases leading to tragic results.

Scott begins the book with an example that I think perfectly encapsulates his broader point. He describes the story of the forestry industry in late 18th century Prussia and Saxony. In order to optimize lumber yields, states decided that there was no need to keep the seemingly unordered naturally grown forest. Instead, they could optimize forest growth, planting only the most valuable trees in a grid-like setup to enable easy access. You can probably guess what comes next.

While the managed forests worked well for one generation, soon the trees stopped growing quite as large or even dying before they could be used for lumber. Without the natural habitat they had evolved to survive in, the trees no longer received the nutrients they needed from the soil. Scientists attempted to replicate the essential features of the forest, to provide the trees with the nutrients they needed while maintaining their controlled environment. As Scott describes, “given the fragility of the simplified production forest, the massive outside intervention that was required to establish it – we might call it the administrators’ forest – is increasingly necessary in order to sustain it as well.” Just as we see in countless cases of government intervention, a single intervention ends up requiring even more intervention and government becomes necessary to maintain the system despite being itself the original source of the problem.

Scott summarizes the situation:

“The metaphorical value of this brief account of scientific production forestry is that it illustrates the dangers of dismembering an exceptionally complex and poorly understood set of relations and processes in order to isolate a single element of instrumental value…Everything that interfered with the efficient production of the key commodity was implacably eliminated. Everything that seemed unrelated to efficient production was ignored. Having come to see the forest as a commodity, scientific forestry set about refashioning it as a commodity machine. Utilitarian simplification in the forest was an effective way of maximizing wood production in the short and intermediate term. Ultimately, however, its emphasis on yield and paper profits, its relatively short time horizon, and, above all, the vast array of consequences it had resolutely bracketed came back to haunt it.” (21)

It is hard to read the excerpt above without immediately thinking of other examples where governments have attempted to replace complex natural systems with more intelligible, but far simpler systems. The remainder of the book goes through many such examples, from the design of cities and languages, to the failed communist experiments of the Soviets and the Chinese and many more. Scott picks out four features that his research suggests lead to poor results of state control. They are, in his words

  1. “Administrative ordering of nature and society”
  2. “High modernist ideology”
  3. “Authoritarian state that is willing and able to use the full weight of its coercive power to bring these high-modernist designs into being”
  4. “Prostrate civil society that lacks the capacity to resist these plans”

1, 3, and 4 are pretty straightforward, but 2 deserves some further discussion. By “high modernist ideology,” Scott refers to the belief that scientists and other experts know how to design a society in a more efficient way than ones that develop without any top down intervention. It is the belief that a centralized plan can trump decentralized spontaneous order. Hayek frequently called this attitude “scientism” in his work. Both Scott and Hayek argue that high modernist thinking is too arrogant. Ancient traditions may look backwards to a modern scientist. Customs may seem strange, cultures don’t always make sense.

But it is important to remember that just because you don’t understand the reason behind something doesn’t mean there isn’t one. To the central planner who only cared about lumber production, natural forests seemed incredibly inefficient. So the solution is simple. Cut out everything we don’t need and just keep the good stuff. In doing so, however, those seemingly useless features often reveal themselves to be essential.

Overall, I found the book to be full of interesting historical examples that each serve to illustrate this theme again and again. One point that I did wish had gotten more attention was the role of corporations and their similar top down nature. Scott briefly mentions that “large-scale capitalism is just as much an agency of homogenization, uniformity, grids, and heroic simplification as the state is.” He is quick to note that there is a major difference that “for capitalists, simplification must pay.” Still, in cases like the German forests, if the results were profitable in the short run and the problems only observable after dozens of years, it is easy to imagine capitalist firms falling into the same trap. I would have liked to see some examples of historical corporations that have also failed to simplify complex systems, but I guess that would require a much longer book. As it stands, the book serves as a useful warning for any attempt to improve a natural process that is not fully understood. Well worth the read.